c1ccccc1

10 ms·

61.
by c1ccccc1 4y ago
My responses: > We're ahead of even the most optimistic timelines — what's happening with transformers is taking everyone by surprise. I admit that I was surprised by how good GPT-2 was. But after that, you expect a lot of effo
62.
by c1ccccc1 4y ago
If the context window is finite, then LLMs actually are Markov chains. It's just that they're a much more efficient way of representing transition probabilities than storing them all in a giant lookup table.
63.
by c1ccccc1 4y ago
I agree with your second point that having deontological boundaries is important. Interestingly, people like Eliezer Yudkowsky (generally respected in the EA movement, I'm pretty sure?) also agree with you about that. He's been re
64.
by c1ccccc1 4y ago
Interestingly, the sun would kill us much sooner than that: in less than 1 billion years the sun's increased output will heat up the Earth's surface to the point of becoming unlivable. Related to the topics of AI apocalypse and pu
65.
by c1ccccc1 4y ago
Depends what Aaronson meant by "being ahead morally"; it's unclear whether he is referring to morality in the sense of being virtuous, or in the sense of making other people better off. Hypothetical Charitable Investor's
66.
by c1ccccc1 4y ago
Let me push back on the notion that there are alternative explanations besides increased concentration of atmospheric CO2 and other greenhouse gases for the current rapid changes in climate. It's one thing to say that we should distrus
67.
by c1ccccc1 4y ago
Thanks for replying with details. The difference for liquid water is small, but for ice it looks like it's actually about 7%, which could be significant. Looks like what's going on is that the constants used are for different temp
68.
by c1ccccc1 4y ago
Is it possible for you to link any of the climate models where you've observed this and point out some of the inaccuracies? (Or to link to an existing blog post or comment that does so?)
69.
by c1ccccc1 4y ago
What do you make of the argument that if you give the police your phone, you are in fact fully cooperating by giving them all data on that phone, and it's not your fault if they don't have sufficient information to interpret that
70.
by c1ccccc1 4y ago
Sorry for what's probably a stupid question: I see there's 2 featured courses on the home page, how do I view other courses? I don't see a search bar anywhere.
71.
by c1ccccc1 4y ago
I'm not the person you replied to, but just wanted to add something to help answer Q2. Communicating information to a large fraction of the populace is generally pretty hard because there are so many memes competing for attention, and
72.
by c1ccccc1 4y ago
I like this explanation and the one in the parent comment quite a bit. Rather than a grand thesis about internet / social media [0] / polarization that one's likely to see in a think-piece, it's the little things about h
73.
by c1ccccc1 4y ago
I agree that's it's possible to create objective collapse theories that are arbitrarily difficult to falsify, but the difference between objective collapse and regular old decoherence due to interaction with the environment has ex
74.
by c1ccccc1 4y ago
Many worlds is absolutely testable, since if we observe collapse in even a single one of these experiments then that completely falsifies many worlds. If one of these experiments discussed in the article had actually observed a collapse, th
75.
by c1ccccc1 4y ago
Wait, I'm pretty sure that if these measurements had shown a collapse, that would have to be taken as disproving many worlds. You can't very well have many worlds when your wavefunction is physically collapsing all the time before
76.
by c1ccccc1 4y ago
It's not just a relabelling. In collapse theories, the wavefunction stops obeying the Schrodinger equation for a moment and discontinuously jumps to a new state. The times when it performs a discontinuous jump are called "measurem
77.
Coxeter and Dynkin Diagrams [pdf](math.ucr.edu)
1 points by c1ccccc1 4y ago|0 comments
78.
by c1ccccc1 4y ago
Is it just me, or is this article really spammy and low-quality? It's full of buzzwords like "advanced knowledge in cutting-edge Machine Learning solutions" and "completely new paradigms for enhanced creativity". Me
79.
by c1ccccc1 4y ago
Sounds like something we could check empirically. Some cosmic rays are protons from outside of the solar system. We could capture those and compare them with local protons to see if they're actually the same kind of particle. (We can c
80.
by c1ccccc1 4y ago
Is that really true about Feynman? I thought he wrote in "Surely You're Joking" that he didn't take psychoactive drugs because he loved thinking and he "didn't want to mess up the machine".
81.
by c1ccccc1 4y ago
Reminds me of this story: https://slatestarcodex.com/2017/11/09/ars-longa-vita-brevis/
82.
by c1ccccc1 4y ago
Fortunately, it should be possible to determine which is true, since there is a very large control group in the form of people who have opted not to be vaccinated.
83.
by c1ccccc1 4y ago
You got the direction backwards. Increase in entropy means that the farther back you go in time, the less information the universe has in it in some sense. The initial state of the universe would therefore be very simple, having low entropy
84.
by c1ccccc1 4y ago
Your position/speed/acceleration analogy is misleading from a physics perspective. The amount of GHG in the atmosphere determines an equilibrium point. More GHG corresponds to a hotter equilibrium. There's some delay in movin
85.
by c1ccccc1 4y ago
> Parthenogenetic offspring in species that use either the XY or the X0 sex-determination system have two X chromosomes and are female. From this we can infer that Jesus had two X chromosomes, and was therefore a trans man.
86.
by c1ccccc1 4y ago
And in both cases those making the claims have somehow managed to ignore existing biological counterexamples.
87.
by c1ccccc1 4y ago
I haven't checked the geological suitability, but assuming you're right about that, putting the waste in those places seems perfectly fine to me.
88.
On black holes, holography, fully homomorphic encryption, etc.(scottaaronson.blog)
11 points by c1ccccc1 4y ago|1 comments
89.
by c1ccccc1 4y ago
That seems super unlikely to be a problem. For long term storage, I'd expect that we'll dig a deep shaft, or use an existing abandoned mine, and then seal it back up. Any group of humans who are willing and able to make a hole thr
90.
by c1ccccc1 4y ago
200,000 years is not much on geological timescales. Here's one example of a place with very stable geology: https://en.wikipedia.org/wiki/Canadian_Shield The other main thing to remember when burying nuclear waste

More ›